2023-03-16

Machine Status

Overall, the machine seems to be in a very good state.

Vacuum is 3.0e-7 Pa (yesterday was 3.2e-7).

TORA signal is greater than around ±100 mV at injection.

FAB signal is significantly larger than this, although at the time-of-writing it's unclear if this is RF noise or beam induced signal.

Preliminary Investigation

We performed several measurements and procedures quickly, without taking long sweeps or many repeats to understand the situation. 

Schottky Measurement 1

With TORA amplifier. Aluminium shield applied over the top of the amplifier. Scope vertical range set to 50 mV/ div because with 20 mV/ div the trace clipped without beam. Channel 1 was set to 50 Ohm, 200 MHz bandwidth and the 80 MHz low-pass filter. Acquisition speed is 800 ps/point (1.25 GSa/s).

Bunched beam signal at injection is completely out of the vertical range of the scope; more than ±200 mV. Initial unbunched data, with a delay of 77 ms, shows peaks above the noise at harmonics up to 8; higher harmonics not checked and noise signal on harmonics 6 and 7 made the signal impossible to see. Harmonic 8 is clearly above the beam-off data. The scaled widths of harmonics 4, 5 and 8 look about the same (as expected). 

Noise Minimisation

We changed back from the TORA amplifier to the original amplifier. We added aluminium shielding around the amplifier and the noise reduced. Furthermore, we expected some induced signal, so wrapped the coaxial cable around a ferrite ring (material 43); this also observably reduced noise so was left in-position.

The reduction in noise is most clear on the large 18 MHz peak.  

Increasing Horizontal Scale

We were previously sampling at 1.25 GSa/s. We wanted to improve the frequency resolution of the peaks, so wanted to increase the total amount of time acquired, but memory is already maximised, so the sample rate would need to be reduced. Reducing the sample rate is expected to increase noise, and we should stay significantly above 160 MSa/s, or we'll alias signal in the pass-band of the (3rd order Butterworth) 80 MHz low-pass filter. 

We increased to a total window is 50 ms, with a sample rate of 500 MSa/s (> 6x 80 MHz). We observed a small increase in noise so decided not to go further, but the resulting peaks were still very clear.

Abrupt Debunch (4kV)

To verify that these peaks are really Schottky peaks, we did a rapid debunch and expected to see a lower and wider peak. A significantly wider peak was observed on multiple harmonics. The wider peak has a worse signal-to-noise ratio, but it is still clearly observable over the no-beam case.

Abrupt Debunch (1kV)

It is difficult to quantitatively compare the adiabatic and 4 kV cases, so a 1 kV rapid-debunch case was performed too. A clear change in the peak width is visible.

The ratio of the widths is around 1.8, but since dP/P is expected to vary as V^(1/4) we expected a change of 1.4. This discrepancy has been explained by Uesugi-san, who has informed us that there is a quadratic dependence between voltage at the power supply and the voltage at the cavity; 4kV and 1kV are at the power supply. Uesugi-san has also recorded the voltage at the cavity so we can use the correct voltages when a more detailed analysis is performed.

dP/P Measurement

We are now confident that the peaks are varying as expected, so will begin taking more systematic data. To start with, we will record abrupt-debunch data for voltage settings 1 kV, 2 kV, 3 kV and 4 kV. We will record this at a delay of 90 ms and 290 ms from injection. We acquired 5 beam-on datasets for each setting and 2 beam-off datasets.

RF voltage at end of ɸ_s sweep

As described by David K. We will record this at a delay of 90 ms and 290 ms from injection. We acquired 5 beam-on datasets for each setting and 2 beam-off datasets. Intermediate voltages were 2.0 kV, 1.8 kV and 2.2 kV.